SkillsHooksPromptsAgentsPersonasModelsPoliciesToolsTemplatesBundlesCategoriesStart here
← Models
Mo modelmultimodalaistable

deepseek-ocr

DeepSeek-OCR is a vision-language model that can perform token-efficient OCR.

id model/deepseek-ocrvendor DeepSeekfamily DeepSeek
Sizes
3b
Capabilities
vision
Model card
https://ollama.com/library/deepseek-ocr ↗

Providers

ProviderModel idGet itTagsPullsUpdated
Ollama ↗deepseek-ocrollama pull deepseek-ocr3525.8K2025-11-19

Same family

ollamalocalvision
Author DeepSeek. Source ollama.com/library (original ↗). License unknown — the source states none; review before redistributing. Imported 2026-09-03. Description, capabilities, sizes, pulls, tag count and updated date as shown on the Ollama library page. Vendor, family and task are inferred from the model name; weights license is not published on the listing.